C / C++ / MFC

Re: How to create a window in general? [modified]

28-Dec-06 10:09

Greetings:

I am trying to figure out how to create a very general window in MFC out of thin air.

Without getting into the grim details, I have a simple MFC application - no Doc/View. When the app launches, a naked CMainFrame is displayed with no child window over it. It is colored with COLOR_APPWORKSPACE. This is the way we wanted it for a variety of reasons.

Now, there is a menu item that must create a very simple window - like a minimal CView - and displays it in the Main Frame as a child window. I do NOT want to maximize this new window to fit the entire client area of the main frame. I would like the new window to have boarders and a caption and the option to maximize but not much else.

I cannot seem to make this happen under MFC. Can anybody help?
Thank you in advance.
Mark

Mark Salsbery28-Dec-06 11:24

28-Dec-06 11:24

MFC windows are created in two steps - one for the C++ CWnd object and one for the Windows HWND.

// step 1
CWnd *pWnd = new CWnd();
// step 2  
pWnd->Create(NULL, _T("Window Name"), WS_CHILD|WS_VISIBLE|WS_OVERLAPPEDWINDOW, CRect(10,10,100,100), 
                pParentWnd, ID_CHILDWND, NULL);

If you want the window to do something useful you'll probably want to derive your own window class
from CWnd.
If you want to use extended window styles you can use the CreateEx() method.

Destruction of windows is generally two steps as well (CFrameWnd-derived windows are an
exception).

// step 1
pWnd->DestroyWindow();
// step 2  
delete pWnd;

-- modified at 17:48 Thursday 28th December, 2006
Changed the WS_ styles.

Mark Salsbery28-Dec-06 11:47

28-Dec-06 11:47

Try this style in the sample code I posted (I changed it there as well) in the Create() call:

WS_CHILD|WS_VISIBLE|WS_OVERLAPPEDWINDOW

This will help you see the window better Big Grin | :-D

Jethro6328-Dec-06 13:32

28-Dec-06 13:32

Hi Mark:

Thanks for your response. Its working except the window is displaying without a caption. I'm getting only a rectangular frame with no caption and no way to move it around. I AM able to resize it though...

Mark

Mark Salsbery28-Dec-06 13:40

28-Dec-06 13:40

Even with the WS_OVERLAPPEDWINDOW style added?

Can you post the line(s) of code calling Create()?

Mark

Jethro6328-Dec-06 13:59

28-Dec-06 13:59

Hi Mark:

Sure, here it is:

<br />
CRuntimeClass*	pRTC;<br />
pRTC = RUNTIME_CLASS( COneView );<br />
pWindow = (CWnd *)pRTC->CreateObject( ) ;<br />
pWindow->Create( (LPCTSTR)m_strTypeOneView, <br />
                  TEXT("One Little View"), <br />
                  WS_CHILD | WS_VISIBLE | WS_OVERLAPPEDWINDOW, <br />
                  CRect(0, 0, 400, 300), this, 4567 );<br />

The "mstrTypeOneView" string contains the result returned from an earlier call to
AfxRegisterWndClass:

<br />
try<br />
{<br />
	m_strTypeOneView = AfxRegisterWndClass(<br />
		0,<br />
		::LoadCursor(NULL, IDC_CROSS),<br />
		(HBRUSH)reinterpret_cast<HBRUSH>(COLOR_BACKGROUND+1),<br />
		::LoadIcon(NULL, IDI_APPLICATION));<br />
}<br />
catch (CResourceException* pEx)<br />
{<br />
	AfxMessageBox(_T("Couldn't register class! (Already registered?)"));<br />
	pEx->Delete();<br />
}<br />

I'm sure that the window registration is working because I have made alterations to the background color and the ICON and the window responds as expected.

Thanks,
Mark

Mark Salsbery28-Dec-06 14:19

28-Dec-06 14:19

Jethro63 wrote:
The "mstrTypeOneView" string contains the result returned from an earlier call to
AfxRegisterWndClass:

How much earlier. It's not safe to store the returned pointer. It should be copied into your own
buffer (a CString is easiest).

The title bar isn't hidden under a toolbar is it?
If so, adjust the CRect top and left params.

Mark

Jethro6328-Dec-06 14:27

28-Dec-06 14:27

Hi Mark:

"m_strTypeOneView" is a CString member of CMainFrame. The call to AfxRegisterWindClass is carried out in CMainFrame::OnCreate. The string returned is copied to "m_strTypeOneView" at that point. So, I believe that the returned class name is protected and not volatile.

I thought of your second point shortly after I sent my last reply to you and moved the upper-left coordinates of the window to 40,40. The window moves but does not change in appearance.

Mark

Mark Salsbery28-Dec-06 14:33

28-Dec-06 14:33

Jethro63 wrote:
"m_strTypeOneView" is a CString member of CMainFrame. The call to AfxRegisterWindClass is carried out in CMainFrame::OnCreate. The string returned is copied to "m_strTypeOneView" at that point. So, I believe that the returned class name is protected and not volatile.

Cool sorry about that Smile | :)

Jethro63 wrote:
I thought of your second point shortly after I sent my last reply to you and moved the upper-left coordinates of the window to 40,40. The window moves but does not change in appearance.

Hmm That's just not cool! What if you create it using the default class (pass NULL for 1st param
of Create())? In my little MFC tester app (that I use to test code I post here) I have this:

pWnd->Create(NULL, _T("Window Name"), WS_CHILD|WS_VISIBLE|WS_OVERLAPPEDWINDOW,
CRect(50,50,300,200), this, 12345, NULL);

And it creates as intended - caption, system menu, max/minimize buttons, etc.

Mark Salsbery28-Dec-06 14:38

28-Dec-06 14:38

CRuntimeClass* pRTC;
pRTC = RUNTIME_CLASS( COneView );
pWindow = (CWnd *)pRTC->CreateObject( ) ;

I haven't looked at the MFC source code in a while but if COneView is CView-derived then maybe
some style flags are getting stripped off by MFC since views are meant to be used in a frame.

Jethro6328-Dec-06 14:45

28-Dec-06 14:45

I think you might be right about that and I am going to quickly try to create a different view derived directly from CWnd.

Unfortunately, I have to go and put my son to bed now, so I will not be able to continue this thread this evening (unless much later).

Thanks again for your advice. I will let you know what happens next.
Mark

prasad_som28-Dec-06 19:41

prasad_som

28-Dec-06 19:41

Jethro63 wrote:
I am going to quickly try to create a different view derived directly from CWnd.

You may like to derive it from CView instead of CWnd.

Prasad
Notifier using ATL | Operator new[],delete[][^]

Jethro6329-Dec-06 3:20

29-Dec-06 3:20

Hi Prasad.

Thank you for your response.

The original window that I am having trouble creating was derived from CView. The suggestion that Mark Salsbery is making is that this could be the problem. I am going to try creating a window derived from CWnd, just to see what happens.

Cheers,
Mark

prasad_som29-Dec-06 7:02

prasad_som

29-Dec-06 7:02

Jethro63 wrote:
The suggestion that Mark Salsbery is making is that this could be the problem

Mark was refering to certain styles like WS_OVERLAPPEDWINDOW, would be inappropriate for view class.

Prasad
Notifier using ATL | Operator new[],delete[][^]

octal GUID convertion to string

prasad_som29-Dec-06 7:08

prasad_som

29-Dec-06 7:08

Probably, this is last post this year. Its friday 11:37 PM here. Can continue on your problem on tuesday.
Happy new year. Smile | :)

Prasad
Notifier using ATL | Operator new[],delete[][^]

vga228-Dec-06 6:43

vga2

28-Dec-06 6:43

Hi
I am using LDAP protocol (OpenLDAP) to speak to a MS active directory server. I have a DN of a user, and wants to retreive this users msExchMailboxGUID attribute which is (as far as I can read) a octal 16 byte array.

After a little research, I discovered a way to convert this value to a GUID that looks like this (string representation)

{A64B2D95-B730-4F2A-8F74-C6A37C11B7E5} (38 characters)

This works now on my computer which is an Intel based computer WinXP 32bit.

Which is what I am used to when I view GUID entries in MS SQL or .NET language.

I read somewhere that the GUID in memory is presented as

DWORD-WORD-WORD-BYTE BYTE-BYTE BYTE BYTE BYTE BYTE BYTE

I am no expert on the little-endian vs big-endian, but to obtain the guid string above, I had to start with byte [3], byte[2], byte[1] and byte[0], to get the first DWORD correct.

Is there anyone who has any advice on how to do this easier, or any knowledge if the first 4 bytes will always be in this order when I use LDAP protocol?

How can I in c++, convert a 4 byte "number" to an ascii string of 8 hex characters?

I did it by using _snprintf, and %02X for each byte, but I had to pick the bytes in the correct order out of the 16 byte array to get the right GUID.

I'm just posting to see if anyone has some advice or comments.

Thank you for reading.

Re: octal GUID convertion to string

Michael Dunn28-Dec-06 7:15

Michael Dunn

28-Dec-06 7:15

Use StringFromCLSID() and CLSIDFromString() to convert between the two formats.

--Mike--
Visual C++ MVP Cool | :cool:

LINKS~! Ericahist | PimpFish | CP SearchBar v3.0 | C++ Forum FAQ

Ford, what's this fish doing in my ear?

Re: octal GUID convertion to string

vga229-Dec-06 2:49

vga2

29-Dec-06 2:49

Hi again,

Thanks for the advice. I didn't get StringFromCLSID to work, as I didn't quite understand what to place into the function, but for my need, I could just copy the 16 octet bytes onto a GUID struct and call StringFromGUID2, and it worked.

Cheers.

Bizarre printf behaviour

Ed.Poore28-Dec-06 6:12

Ed.Poore

28-Dec-06 6:12

My Dad just came across the following "interesting" bit of code while trying something out:

unsigned char fred[] = {1, 2, 3, 4};
unsigned char *p = fred;

printf("%d %d %d %d\n", fred[0], fred[1], fred[2], fred[3]);
// outputs: 1 2 3 4
printf("%d %d %d %d\n", *p++, *p++, *p++, *p++);
// outputs: 4 3 2 1

Looks like printf is being evaluated backwards, any idea as to the reasoning for this? Happens on the compiler he was using for the M16C and also on MSVC8.

I have no idea what I just said. But my intentions were sincere.

David Crow28-Dec-06 6:29

David Crow

28-Dec-06 6:29

Ed.Poore wrote:
...any idea as to the reasoning for this?

The result is undefined because the order of evaluation of function arguments are undefined. For more on this, search for sequence points.

"Approved Workmen Are Not Ashamed" - 2 Timothy 2:15

"Judge not by the eye but by the heart." - Native American Proverb

Ed.Poore28-Dec-06 6:57

Ed.Poore

28-Dec-06 6:57

Bizarre to say the least but at least we know it's there and can avoid it. Beer | [beer]

I have no idea what I just said. But my intentions were sincere.

Mark Salsbery28-Dec-06 7:10

28-Dec-06 7:10

There's nothing bizarre about it. If the arguments are pushed on the stack from right to left
then it's behaving as implemented.

As David stated, this behavior is undefined. A C compiler is free to implement how function
calls are made in any way so relying on a specific behavior like this would be bad Smile | :)

Mark